Daksh Soni

Architected and maintained multi-agent LLM systems for RFP/RFI automation, finance SKU analysis, and a supply chain Marvis chatbot, improving operational efficiency by 40–60%. Reduced RFP/RFI drafting time by 40% (saved 120+ man-hours/month) and increased pricing accuracy by 30% while cutting analysis time from 15 to 6 hours per cycle.

Founded an LLM-driven personalized learning platform, deployed to 200+ students across 3 schools, improving engagement by 35% and retention by 28%. Built end-to-end data extraction and LLM personalization pipelines and led architecture/prototyping and beta deployment with cross-functional stakeholders.

Developed an enterprise intranet chatbot using RAG, rerankers, and guardrails for 500+ employees, increasing response relevance by 20%. Automated CI/CD with GitLab and Airflow (reducing release effort by 30%) and created an evaluation-focused testing framework that reduced production bugs by 45%.

Designed an LLM-based meeting summarization system, cutting documentation effort by 50% (about 30 minutes saved per meeting). Implemented a resource optimization proof-of-concept using skill-graph modeling to reduce staffing planning time by 40%.

Pursued a B.Tech in Computer Science Engineering with a minor in Big Data from 2020 to 2024.